T H E STEREOCHEMISTRY AND VALENCE STATES OF NICKEL
R. S. NYHOLM School of Chemistry, N e w South Wales University of Technology, Sydney, Australia Received March 6 , 1963 CONTENTS
I. Introduction.. . . . . . . . . . . . . . . ................................ A . General. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Bond types and stereoche ........... 11. The bivalent state of nickel., . . . . . . . . . . . . . . A. Ionic lattices.,. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Four-covalent complexes of nickel.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Paramagnetic complexes. . . . . . . . . . . . . . . . . . . . . 2. Diamagnetic complexes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........... 3 . Factors influencing the nickel-ligand bond type 4 . Effect of extramolecular environment of the molecule (or complex ion) on the stereochemistry of the nickel a t o m . . ......................... C. Five-covalent complexes. . . . . . . . . . . . . . . . . . . . .......... D. Six-covalent complexes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Paramagnetic six-covalent complexes, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .......... 2 . Diamagnetic six-covalent complexes.. . . . . 111. The tervalent state of nickel.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Five-covalent complexes.. . ............ B. Six-covalent complexes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... IV. The quadrivalent state of nickel ................. ... ... V. The univalent state of nickel.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............................... VI. The zerovalent state of nickel. . . . . . . . . . . . A. Nickel carbonyl. . . . . . . . . . . . . . . . . . . . . . . . . . . . .......... B. Substituted nickel carbonyl compounds, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............................ VII. Summary and conclusions.. . . . VIII. References.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..........
263 253 255 267 267 257 267 273 275
277 278 2 30 2 30 233 234 235 237 290 295 297 297 300 304 304
I. INTRODUCTION A. GENERAL
The metals of the first transition series from titanium to copper are noted for the variable valence which they display in their simple salts and complex compounds. This behavior arises from the presence of an incomplete 3d shell of electrons; the comparative ease of removal of two or more electrons from the metal atom, together with the availability of d orbitals for bond formation, enables these elements to form a large number of complex compounds with various stereochemical arrangements. However, for a long time, nickel was regarded as more conservative than the other transition elements in regard to variable valence. The bivalent state was recognized in the simple salts and in many complex compounds but until quite recently the existence of higher valence states was considered uncertain (175). In nickel tetracarbonyl, known since 1888 (132), the metal wm taken, a t least formally, as zerovalent, but since car263
264
R. S. NYHOLM
bonyls were often regarded as a class of compounds for which the ordinary rules of valence were not necessarily applicable, the valence of the metal remained uncertain. Thus, as recently as 1930, a formula for nickel carbonyl involving a five-membered ring consisting of four carbon atoms and a bivalent nickel atom (formula XII) was still being seriously considered (133, 183). Formulations such as this reflect the reluctance to accept the possibility of valence states other than the common one shown in the simple salts. Also, although oxides which vaiied in composition from Ni203to Ni02 were known, it was considered that these might well be peroxides rather than compounds in which the valence of nickel was greater than 2. In view of the difficulties of deciding the actual valence state of a metal in higher oxides, this caution is quite understandable. However, during the last twenty years, the valence states of 0, 1, 3, and 4 have become firmly accepted, even though some doubt still remains as to whether nickel is really univalent in the compounds having the empirical formulas K2Ni(CN), and NiCN. The valence state of 6 has been claimed also (68), but adequate support for this is lacking. In this review those valencies which have been definitely established will be discussed in detail and those concerning which doubt still exists will be examined. Such stereochemistry of these valence states as is known will be reviewed, with special reference to work published during the past two decades. The reasons for the rapid development in the pure chemistry of nickel during the past twenty years are relevant to this discussion. During this period, inorganic chemistry in general, and coordination chemistry in particular, has received a great stimulus, largely due to three main developments. The first of these is the application of quantum mechanics to the study of chemical problems. Although the theoretical investigators are limited in the main to purely qualitative predictions concerning bond strength and stereochemistry, nevertheless the results have enabled chemists to understand the nature of the chemical bond and the factors which influence its strength and directional properties. Furthermore, the ability of the theoretical workers to predict what stereochemical arrangements are energetically feasible, in terms of the bond orbitals used, has led to notable advances. The fact that some controversy developed over certain predictions, eg., with regard to the square arrangement in bivalent palladium, bivalent platinum, and certain bivalent nickel complexes, has been in itself most stimulating, since much experimental work has followed in attempts to support or disprove the theoretical predictions. Not the least valuable outcome of this work has been the light which has been thrown on the nature of the groups required for the stabilization of unusual valence states. The second influence at work has been the application of various physical methods to inorganic chemistry. Many of these methods are not “new” in the sense of having been recently discovered, but they could be applied with a new sense of purpose following the enunciation of the relationships esisting between bonding orbitals and stereochemistry as laid down by the workers in quantum mechanics; in a way, many of these methods are complementary with quantum mechanics. Thus, a magnetic moment of itself tells one nothing about the stereochemistry of nickel unless some theory relating stereochemistry with the num-
STEREOCHEMISTRY AND VALENCE STATES O F NICKEL
265
ber of unpaired electrons is acceptable. By the same token, quantum mechanics is often able to indicate that certain stereochemical arrangements only are energetically feasible but it requires some physical measurement to decide between them. The application of some of these physical techniques has encouraged the use of what were, until recently, uncommon ligands. This arises from the fact that suitable physical properties must be possessed by a molecule before certain techniques are applicable; thus, benzene-soluble non-electrolytes are desirable for measurements of electric dipole moment and the insolubility of metal-ammines has resulted in the greater use of organic arsines, phosphines, and of disubstituted sulfides for this purpose. As will be seen presently, ligands with donor atoms such as phosphorus, arsenic, and sulfur have proven of great value for the stabilization of Ni(O), K(III), and Ni(1V). The third factor which has resulted in considerable advances in inorganic chemistry is the great development in fluorine chemistry. Owing to its very high electron affinity, fluorine is an ideal element for the removal of electrons from metals and hence for the preparation of metals in higher valence states. The use of elemental fluorine and bromine trifluoride has resulted in the stabilization of several higher valencies of metals of the first transition series: e.g., Co(1V) (105), Cu(II1) (88), and Ni(1V) (88, 105) as the complex KzXiF6. B. BOND TYPES LWD STEREOCHEMISTRY
The term “stereochemistry,” as generally understood, is used in reference to the arrangement of the bonds around a particular atom. Some qualification is necessary, however, because two extreme types of binding are recognized in inorganic compounds: covalent (e.g., SnC14) and electrovalent (e.g., CsF). In compounds in which the first type of binding occurs, the arrangement of the bonds around the metal atom depends upon the particular atomic orbitals of the metal atom which are used for bond formation. In the stannic chloride molecule, for example, four equivalent 5s5p3 orbitals of the tin atom overlap with the 2p orbitals of four chlorine atoms to form four u bonds which are arranged tetrahedrally. The arrangement is dependent upon the number and nature of the atomic orbitals used ; the stereochemical arrangements resulting from the various combinations of orbitals giving coordination numbers from 2 to 8 have been summarized by G. E. Kimball (102). In electrovalent compounds, however, the whole crystal is one giant molecule and the arrangement of anions about the cation is decided by the relative sizes, relative numbers, charges, and polarizabilities of the ions involved. For electrovalent compounds it is not necessary to consider the orbitals of the metal atom, except in those cases where resonance may occur between the purely ionic structure and a possible covalent structure which has the same directional properties as these resulting from the purely electrostatic model; this is, for example, quite feasible in certain octahedral and tetrahedral complexes. As used here, the term “stereochemistry” implies the arrangement around the metal atom of covalent bonds which are considered to use definite orbitals of the metal atom. For the sake of completeness, however, the crystal structures of the more common ionic lattices of bivalent nickel compounds which are known will be mentioned. Inevitably there
266
R . S . NI?IOLM
are compounds which appear to be intermediate between these two classes. Thus, potassium ferrioxalate is a molecule in which the six iron-oxygen bonds are arranged octahedrally around the ferric atom but the magnetic moment (5.9 Bohr magnetons) clearly indicates that the five 31 orbitals are singly occupied and hence that none is available for covalent-bond formation with the oxygen atoms. Since the octahedral arrangement in covalent complexes of transition metals was considered to arise from the use of six 3d24sip3orbitals, L. Pauling (153) proposed that the bonds in this molecule are essentially “ionic.” The term has been criticized for various reasons, the chief objection being that the physical properties and stability of these compounds, together with the directional properties of the iron-oxygen bonds in this and in many other molecules, as in ferric tris(acetylacetone), warrants the conception that the bond is “covalent” in the simple sense in which this term is generally used. Later Pauling (157, 159) suggested that the four 4s4p3 bonds might resonate among the six positions, thereby giving to each bond a certain covalent character; however, this picture is not very satisfactory in view of the difference between the angles a t the metal atom in a tetrahedral and in an octahedral complex. More recently it has been proposed by H. Taube (187) that the bonds in a compound such as K3Fe(Cz0& should be regarded as covalent but using the “outer” 4d orbitals of the metal atom, involving the combination 4sip34d2.In K3Fe(CN)6,on the other hand, the “inner” 3d orbitals are used in the hybridization 3d24s4p3.A similar suggestion was made by F. H. Burstall and R. S. Nyholm (27) to explain the magnetic properties of the [Ni(dipyridyl)j]++and similar transition metal ions. The possibility that these “outer” 4d orbitals were utilized in bond formation, a t least in square complexes, had been proposed some time ago by M. L. Huggins (go), who suggested that the hybridization in square cupric complexes involved 4s4p24d bonds rather than 3d4s4p2 orbitals; by this combination the necessity for the promotion of an electron (to a 4p orbital) is avoided. For square complexes essentially the same idea has been advanced (170) by P. R$y and D. N. Sen, who proposed that cupric complexes of both 3d4s4p2and 4s4p24dtypes may exist. Theoretical support from calculations of overlap integrals for the use of 4d orbitals by metals of the first transition series for both u- and n-bond formation has recently been obtained (43, 209). Calculations of overlap integrals have established that, a t bond lengths of chemical interest, the strength of octahedral 4s4p34d2bonds, using the outer d orbitals, is greater than 4s4p3 bonds a t the same length. Furthermore, it has been found that the more electronegative ligands should favor the use of the “outer” 4d orbitals, because these 4d orbitals project much more than do the 3d orbitals, thus enabling overlap a t greater distances. Such bonds clearly will possess a considerable polarity, since the region of the overlap, i.e., maximum electron density, is closer to the ligand than is the case with bonds to groups of lower electronegativity. Thus the descriptions of the bonds in complexes in which no electron pairing occurs as “ionic” or “using higher level covalent bond orbitals’’ do not differ as much as might appear a t first sight. In this article the term “ionic” will be used to mean that, as deduced from
STEREOCHEMISTRY A S D VALEKCE STATES O F RICKEL
26 7
magnetic measurements, the number of unpaired electrons in a complex is the same as in the free ion and hence that orbitals used for bond formation are above the 3d shell. In general these will be 4s4p3 or 4s4p24d in four-covalent complexes and 4sip34d2in six-covalent complexes. Here the term “covalent complex” will be used in reference to those complexes which show maximum electron pairing and hence which have 3d (“inner” d ) orbitals available for bond formation. It has been known empirically for some time that electron pairing in transition complexes, and hence the use of lower 3d orbitals for bond formation, is favored by the less electronegative groups, e.g., -CN, --KOz, -SR. Very electronegative groups like fluorine can form complexes with the first transition series, e.g., K3FeF6,KOCoF6,but electron pairing occurs in only one of these, the compound K2NiF6 (88, 143).
11. THE BIVALENTSTATEOF NICKEL The stereochemistry of nickel is most conveniently discussed under the various valence states; of these the common bivalent compounds will be considered first. We shall then discuss the higher and lower valence states in that order. A. IONIC LATTICES
Although not falling strictly within the stated purview of this article, arrangements found in compounds of bivalent nickel with ionic lattices are of interest. Nickel chloride and nickel iodide both crystallize in the cadmium chloride type lattice, while nickel bromide has the cadmium iodide type of structure (204). These two structures are very similar but differ in the manner in which the layers pack together. The halogen ions are close-packed in a cubic manner in the cadmium chloride lattice but hexagonal in the cadmium iodide lattice. Nickel fluoride, however, crystallizes with a rutile structure; each nickel ion is surrounded by six equidistant fluoride ions which are arranged a t the corners of an octahedron (203). Nickel oxide has a simple body-centered sodium chloride type lattice, but the sulfide possesses a peculiar 5 :5 coordinated structure. The crystal structures of the anhydrous sulfate, nitrate, and other oxyanion salts do not appear to have been studied. The structures of the above salts have been discussed in detail by A. F. Wells (205). Irrespective of the anion, hepta- and hexahydrated nickel salts are six-covalent complexes containing the [Ni(H20)#+cation. In its covalent complexes and in complex ions, bivalent nickel is usually fouror six-covalent. In both cases the compounds may be divided into two classes on the basis of the magnetic moments. B. FOUR-COVALENT COMPLEXES O F NICKEL
1. Paramagnetic complexes
Complexes in this class are generally green or blue in color, in sharp contrast with the diamagnetic complexes, which are usually red, brown, or yellow. Typical paradmagnetic complexes are bis(acety1acetone) nickel and bis(triethy1phosphine) nickel nitrate. The magnetic moments of these compounds vary between 3.2 and 3.4 Bohr magnetons (101), indicating two unpaired electrons.
268
R. S. NYHOLM
The calculated magnetic moment for two unpaired electrons on the “spin only” formula ( p = d n ( n 2 ) ) ’ is 2.83 Bohr magnetons; this simple formula is commonly adopted as a first approximation for elements of the first transition series. In practice, however, the moments generally exceed slightly the “spin only” value unless the atom is in an S state, Le., has no resultant orbital angular momentum; Fe(II1) and Mn(II), each with five unpaired electrons, are examples
+
TABLE 1 Electronic configuration of nickel complexes
3d
4s
NPAIUED ELEC-
CALCULATED
TRONS
(SPIN ONLY)
fi (B.M.)
40
Metallic nickel (ground state)
Ferromagnetic
NiO (tetrahedral 4 8 4 ~ 9bonds)’
Diamagnetic
un
Ni+,+ (free ion)
2.83
Ni” (tetrahedral 4 8 4 ~ :bonds)
2
2.83
Ni’I (square planar 3d484pt bonds)
0
Diamagnetic
Ni” (octahedral 4s4pr4d’ bonds)
2
2. &3
Ni’I (octahedral 3d2484ps bonds)
0
Diamagnetict
3
3.88
Nix11 (square pyramid 3d484pa bonds)
1
1.73
NiIII (square pyramid 3d*484pt bonds)
1
1.73
NiIII (octahedral 3d9484p: bonds)
1
1.73
4
4.90
0
Diamagnetic
Ni++t (free ion)
Ni(free ion)
Nix’ (octahedral 3dz4s4p’ bonds)
-.
~
Bonding orbitals shown thus: f Assuming that promotion of two electrons to a 5s orbital OCCUR. If the promotion were to 4d orbitals, paramagnetum with two unpaired electrons would be expected.
of atoms in S states. This small excess, which we shall call for convenience the “orbital increment,” arises from incomplete quenching of the orbital contribution to the moment (200). This will be discussed again presently. PyTotv, since the two electrons in paramagnetic Ni(I1) complexes presumably occupy 3d orbitals (table l),no 3d orbitals are left for bond formation. The immediately available 1n
is the number of unpaired electrons,
STEREOCHEMISTRY AND VALENCE STATES O F NICKEL
269
orbitals above the 3 d level for use in a four-covalent complex are the 4s4pa orbitals; assuming that these 4s4p3 orbitals are used, L. Pauling (151, 152) predicted that paramagnetic four-covalent complexes would be tetrahedral. This assumption has been widely accepted, but experimental support for the hypothesis has only recently been forthcoming. As recently as 1950 N. V. Sidgwick (178) pointed out that a t that time no direct experimental support for the assumption was available. The evidence in support of this hypothesis may be summarized as follows: (a) Theoretical: Since the 4s4p3 orbitals are the first four empty orbitals above the filled 3d shell, it is reasonable, a t first sight, to assume that these will be the ones used for bond formation in a four-covalent complex. Such a simple assumption has been proven valid for the elements of the first two rows of the Periodic Table (e.g., four-covalent nitrogen and phosphorus; four-covalent carbon and silicon; the four-covalent compleses of Zn(II), Hg(II), Cd(II), and Cu(1)). However, this argument loses much of its force when one considers the stereochemistry of four-covalent complexes of bivalent copper, for which a somewhat similar electronic arrangement occurs ( 3 d 9 ) , all five 3d orbitals being unavailable for bond formation. Yet in cupric compounds the arrangement is invariably square planar, an arrangement requiring the use of dsp2 orbitals. For any particular transition element, quantum-mechanical calculations cannot yet be carried out with the accuracy necessary to decide whether the 3d4s4p2, 4s4p24d square, or the 4s4p3 tetrahedral arrangement will have the lower energy. ( b ) Electric dipole moments: K. A. Jensen (95) examined the complex Ni(N03)2.2(CZH6)3P and concluded that the nickel atom must be four-covalent because the compound is monomeric in freezing benzene. The high electric dipole moment (8.85 D) clearly excludes the trans-planar arrangement and is compatible only with a tetrahedral or a cis-planar molecule. Until recently the trans-arrangement mas generally considered to be the thermodynamically more stable form in planar complexes; hence the cis-planar arrangement was regarded as improbable. However, it is known (36) that in the case of planar bivalent platinum complexes the cis-form is often the isomer with the greater bond energy. The equilibrium cis trans has been studied in benzene solution. Care has to be taken when interpreting equilibria in solution, because the position of the equilibrium cis trans in a planar complex is affected both by the relative bond energies of the two isomers and by the entropy changes involved. The expression for such equilibria
*
*
-RT In K p = AG
=
AH - TAX
makes this clear. Furthermore, with Pt(I1) complexes the cis-arrangement appears to be formed most readily when the anion is most electronegative and when the ligand has a high capacity for double-bond formation. Again, since in square complexes ?r binding is favored for two bonds a t right angles (43), the cis-arrangement could quite easily be the more stable for this reason. Thus with Pt(I1) it is found that the cis-arrangement appears to be favored by C1 > Br > I as anion and by those ligands which, whilst having a weaker
270
R. S. NYIIOLM
capacity for a-bond formation, may be expected to form strong a bonds, e.g., antimony, tellurium (97, 137). To sum up, the dipole moment indicates either the tetrahedral arrangement or the cis-planar arrangement. (c) Size of the orbital contribution to the magnetic moment: The effect of stereochemistry upon the size of the orbital contribution to the magnetic moment of transition metal complexes has been studied by several workers during the past few years. Although the magnetic moments of complexes of the first transition series approximate to the “spin only” formula, they may sometimes exceed the calculated value by as much as 1.5 Bohr magnetons, e.g., with certain octahedral Co(I1) complexes. This orbital increment is generally attributed to incomplete quenching of orbital motion by the electrical fields created by the attached atoms. If this quenching is complete a spin only value results, but where this is not so, the size of the orbital increment is an indication of the asymmetry of the field resulting from the attached atoms and hence of the stereochemistry. It has been shown (71, 160, 161, 200) that the “quenching” of the orbital contribution in octahedral nickel complexes should be greater than for those in which the attached groups are tetrahedral. A study of a large number of six-covalent complex ions of the type [Nixs]++,where X is HzO, NH3, %en, xdipy, %‘o-phen (101), reveals that the magnetic moment is little affected by the nature of the above ligands and generally lies between 3.1 and 3.2 Bohr magnetons. Fourcovalent complexes such as [l;i(acetyla~etone)~]~ however, have magnetic moments in the vicinity of 3.3-3.4 Bohr magnetons. This appears to support the tetrahedral hypothesis, but it should be pointed out that no theoretical studies have been reported yet on the effect of a planar arrangement of surrounding atoms upon the size of the orbital contribution in a paramagnetic molecule, i.e., if the bonds were 4s4p24d.At present the size of the orbital contribution may be taken as suggestive of four-as against six-covalence, it does not establish definitely that the arrangement is tetrahedral rather than planar. For a more detailed discussion of this subject, with a survey of recent work, see rc’yholm (142a). ( d ) Fine structure of the K x-ray absorption edge: The fine structure of the K absorption edge is related to the stereochemistry in the following way : When electrons are excited from the K shell of a metal atom, they are lifted to suitable vacant orbitals on the outside of the atom. For the [Cu(HZO)#+ and [Cu(NH3)41+ ions, for example (7), two absorption maxima are observed, one of which the investigators identified as a 1s + 4p transition and the other as an unresolved mixture of 1s + 5 p , Is + Bp, etc., transitions. Clearly an unequivocal assignment of the type of transition gives an indication of the lowest vacant orbital on the outside of the atom, provided the transition to this orbital is not forbidden by the selection rules. Then, if we have, as with Ni(II), the possibility of two types of four-covalent structure in one of which all 4p orbitals are filled, as in tetrahedral 4s4p3 complexes, and in the other a vacant 4p orbital is available, as in square 3d4s4p2 complexes, it should be possible to distinguish between them by observing the peaks in the spectrum. A peak which is attributed to a 1s + 4 p
STEREOCHEMISTRY AND VALENCE STATES O F NICKEL
271
transition has been reported by W. W. Beeman and H. P. Hanson (74) for a series of transition metal complexes which they examined. When the fourcovalent Ni(I1) complexes were investigated, it was found that the diamagnetic complexes show the expected 1s 4 p transition peak, but the paramagnetic complexes do not. Since in the first instance the bond orbitals are 3d4s4p2, a vacant 4 p orbital is available to receive the excited electron, but in the second case the absence of a vacant 4 p orbital is most logically explained by the assumption that all three 4 p orbitals are being used for bond formation, Le., for 4s4p3 bonds. Further work has recently been reported by G. Mitchell and W. W. Beeman (130). The absence of the peak attributed to a Is --+ 4 p transition in bis(8-hydroxyquinoline) nickel(I1) is taken as evidence for a 4s4p3 tetrahedral arrangement. The fact that the peak is also missing in bis(salicyla1dehyde) nickel(I1) dihydrate is taken as evidence for a tetrahedral arrangement here also. However, if this were an octahedral 4s4p34d2complex there would still be no vacant 41, orbital; hence in this case the absence of the peak is as compatible with 4s4p34d2binding as it is with 4s4p3bonds. Hence it is important to establish that the nickel atom is actually four-covalent before attempting to decide the arrangement of the bonds. Work has been done on this subject by E. Vainshtein (195, 196), who found that YiCl?, NiS04, and [Ni(NH3)4]++ions showed the same K absorption edges in both alcoholic and aqueous solutions. The structure of the [Ni(CX)4]-- ion under the same conditions was entirely different. In this case it appears that one is distinguishing between octahedral ions, e.g., [Ni(H20)8]++,and square [Ni(CN),]- - ions. The evidence provided here in regard to four-covalent Ni(I1) is of course indirect but can be taken as supporting the view that the paramagnetic complexes are tetrahedral. ( e ) The Cotton efect: Since the reliability of conclusions as to the stereochemistry of metals on the basis of the Cotton effect has been severely questioned, no weight will be attributed to this evidence here, but a brief summary of the phenomenon is called for. The Cotton effect for metal complexes containing an optically active carbon atom is characterized by three distinct features. A substance which shows this effect has (i) an anomalous rotatory dispersion, (ii) zero rotation a t a wave length in the vicinity of the maximum of the absorption band, and (iii) circular dichroism with maximum ellipticity in the region of zero rotation, i.e., when absorption is a maximum. P. Pfeiffer and his collaborators (164) proposed that for the Cotton effect to occur in a metal complex, the molecule must contain not only an asymmetric carbon atom but also a chromophoric metal atom which is asymmetric, Le., has a tetrahedral arrangement. Since the compound bis(salicyla1dehyde d-propylenediamine) nickel(I1) shows the Cotton effect, it was concluded that the Ni(I1) atom is tetrahedrally coordinated. The inactive form of this compound, however, has been shown by D. P. Mellor (119) to be diamagnetic; hence the Ni(I1) atom is square coordinated. In contrast with this conclusion, H. S. French and G. Corbett (65) have concluded that the Ni(I1) atom is tetrahedrally coordinated in bis(d-formylcamphor) nickel(II), which compound is paramagnetic. Clearly there is some anomaly here if the mag-
272
R.
S. “ H O L M
netic moment can be accepted as a reliable criterion of stereochemistry. D. P. Mellor (119) has examined this problem carefully and points out that in the second compound it is by no means certain that the band in which the Cotton effect occurs can be attributed to the nickel atom. He concludes that the appearance of the Cotton effect is more likely to be concerned with the covalent character of the metal-ligand bonds than with their stereochemistry. An indirect correlation with stereochemistry is implicit here, since strong “covalent” (3d4s4p2) bonds indicate a square arrangement; nevertheless, there is no necessity for recourse to rotatory dispersion measurements to decide this, since the bond character may be ascertained as a rule simply by observation of the absorption spectra alone. More work is needed on the Cotton effect in order to establish with greater certainty its relation to other bond properties, but in the meantime it is considered unsafe to accept conclusions based thereon as reliable evidence for a tetrahedral arrangement of the bonds around a nickel atom. A. E. Martell and M. Calvin (118) also conclude that evidence so far presented does not enable one to decide with any confidence whether the Cotton effect is caused by the presence of an asymmetric metal atom or by coupling between the chromophore and the asymmetric center. (f) X - r a y crystallographic analysis: It is remarkable that until recently no x-ray crystallographic investigation of a paramagnetic four-covalent nickel complex had been carried out. Two such compounds, however, have nom been examined. D. H. Curtiss, F. K. C. Lyle, and E. C. Lingafelter (46) have compared the x-ray powder diagrams of anhydrous bis(salicyla1dehyde) nickel(I1) with those of the corresponding bivalent zinc and copper compounds. They have shown that x-ray powder diagrams of the nickel and zinc compounds are almost identical. However, the cupric complex, which has been shown to be square planar (181), gives an entirely different powder pattern. Keeping in mind the fact that all four-covalent zinc complexes which have been investigated are tetrahedral whilst all four-covalent cupric complexes which have been studied are square planar, the above findings support the view that the nickel compound has a tetrahedral arrangement. Unfortunately the above workers were unable to obtain single crystals of the anhydrous compound, which was obtained as a powder by dehydration of the dihydrate. A similar dihydrate of nickel bis(acetylacetone), presumably octahedral, is known; from this dihydrate, crystals of the anhydrous complex may be obtained by sublimation. An x-ray examination of these crystals is being carried out by G. J. Bullen and K. Lonsdale (25); so far it has been established that the compound is not isomorphous with the corresponding cupric compound, which had been shown previously by x-ray studies t o be square planar (42). The structure determination is incomplete, but the nickel atom is most probably a t the center of a regular tetrahedron of oxygen atoms. For bis(salicyla1dehyde) nickel(II), M. Calvin and N. C. hlelchior (30) have proposed a square planar arrangement of the bonds, because they believe that a 3d orbital is in some way involved in the binding. Resonance between ionic
STEREOCHEMISTRY AND VALENCE STATES OF NICKEL
273
and 3d4s4p2 binding is envisaged, one unpaired electron being in a 3d and one in a 4d orbital. Theoretical work, however, suggests (43) that were a d orbital involved in the binding it would be a 4d and not a 3d, owing to the high electronegativity of the attached groups; the problem is thus most likely to be one of distinguishing between the 4s4p3 and the 4s4p24d combinations, since no promotion of electrons is then required. 2. Diamagnetic complexes
As previously mentioned, compounds of this class usually vary in color from yellow to red or brown. This correlation between diamagnetism and color has been examined by several workers, but several exceptions to the general rule have been observed. For example, the vermillion-colored compound, bis(1-hydroxyacridine) nickel(I1) has a moment of 3.2 Bohr magnetons (131); on the other hand, green bis(formylcamphorethy1enediamine) nickel(I1) is diamagnetic in the solid state (126, 210). D. P. Mellor, J. E. Mills, H. A. McKenzie, and L. N. Short (131) suggested that, since the color of the attached group itself could confuse the issue, the absorption spectrum of the complex would be a more valuable basis for comparison. From the study of the spectra of a large number of four-covalent N(I1) compounds, and of the ligand by itself (in a suitable solvent in each case), it was found that the diamagnetic complexes usually exhibited a sharp absorption band in the vicinity of 4000 i'i. This band is very intense, €-values between 3,000 and 12,000 being reported. The shape of the band is also important, for, unlike the bands of most inorganic complexes, this one is very sharp and closely resembles those observed in atomic spectra. However, there are at least two exceptions to the rule that this band is diagnostic of diamagnetism; thus, the compound bis(formylcamphorethy1enediamine) nickel(II), which is diamagnetic in the solid state, becomes paramagnetic when dissolved in alcohol but still shows the characteristic band near 4000 8. (131, 210). To sum up, this sharp absorption band near 4000 8. and/or diamagnetism may be regarded as diagnostic of a square planar arrangement as a rule but it is not an infallible guide, because a t least one six-covalent Si(1I) complex is known having both of these properties (see page 2 8 3 ) . The conclusion that diamagnetic four-covalent complexes are square planar is well founded on the following pieces of self-consistent evidence: ( a ) Theoretical: The diamagnetism of four-covalent Ni(I1) complexes is most simply interpreted to mean that the eight 3d electrons of an Ni++ ion are forced to pair off, using four fully occupied 3d orbitals. This leaves one 3d orbital empty and when this is hybridized with the next three vacant orbitals the binding mill be 3d4s4pZ (table 1). This results in a square planar arrangement (102). The argument was put forward by L. Pauling in 1931 (151, 152); its subsequent confirmation is one of the notable successes of quantum mechanics. ( b ) Electric d i p o l e momcnts: Follon ing his investigations of the dipole moments of the two isomers of bivalent platinum complexes of the type [Pt(hal)z.2R3P(As)Io (94, 95), I
